Medication-safety programs commonly identify medicines, classes, interactions, or prescribing situations associated with elevated harm. These approaches are clinically useful, but a medicine categorized as high risk does not confer a constant probability of harm across patients, settings, or time. This Current Opinion argues that medication-harm prevention requires a more explicitly patient-level and dynamic interpretation of risk. Intrinsic and regimen-dependent drug hazard should be considered alongside three additional domains: patient susceptibility, monitoring capacity, and rescueability if deterioration occurs. Susceptibility determines how strongly a given exposure may translate into injury; monitoring capacity determines whether an emerging problem is likely to become observable while action remains possible; and rescueability concerns whether recognition can be translated into timely, effective mitigation before consequences become severe or irreversible. These domains can change independently during acute illness, treatment modification, discharge, deterioration in organ function, loss of follow-up, or changes in care setting. The proposed synthesis therefore treats “high risk” as a state that may migrate rather than as a permanent property of a medicine. Existing explicit prescribing criteria, medication-safety tools, and risk-prediction approaches remain important, but they answer different questions and should not be interpreted as interchangeable patient-harm estimates. The proposed model is conceptual rather than validated: its constructs require operational definition, temporal measurement, external validation, and prospective evaluation before clinical implementation. Its principal value is to sharpen how medication-safety priority is reasoned about when the same drug can be relatively manageable in one patient-state yet substantially more hazardous in another.
